Published:

56 comments

From today ConHome is making two changes to how we present opinion polls.

AmendedThe first is presentationally small but we hope instructive. Every time a new opinion poll is published we offer a projection of the likely Tory majority that would emerge from the percentage numbers in the latest survey and also from our polls of polls. Saturday's YouGov poll with a 40% to 31% Tory lead, for example, was projected to produce a Tory majority of 2 according to UK Polling Report's swing calculator. Unusually the Poll of Polls also projected the same majority.

We no longer believe that that tells the full story. Our belief is that a 40% to 31% Tory lead would probably produce a much healthier Tory majority. The UK Polling Report calculator is based on a uniform national swing. It is right for UKPR to stick with the idea of a uniform swing but is such a swing likely?  We think not. We think there is going to be an unwind of anti-Tory tactical voting from previous elections and the emergence of anti-Labour tactical voting. We also expect the extra investment that CCHQ is putting into target seats will pay off. There are also suspicions that the core vote strategy being pursued by Brown will win him more support in heartland Labour constituencies but not among the aspirational swing voters of Middle England once so loved by New Labour. Some of these factors were analysed by academic Quentin Langley in a recent post for ConservativeHome's Platform.

Because it's difficult to know the potency of these factors we are not adjusting the UKPR projection number but we are adding a 'plus' sign to the graphic as a reminder to readers that the majority is likely to be larger. One pollster agreed that the majority could easily be 30 to 40 seats on a 9% Tory advantage.

The other change we are making is to the ConHome Poll of Polls. Up until now we have calculated a mean average of the most recent poll from each of the five biggest pollsters: ComRes, ICM, Ipsos MORI, Populus and YouGov. In this model a MORI poll from one month ago has, say, the same weight as a YouGov poll from one day ago. From now on we will average the five most recent polls as long as they are from those five market researchers or from Angus Reid (now polling for Political Betting) and BPIX (an occasional pollster for the Mail on Sunday). It may be that we will include two polls from, say, YouGov if they have provided two of the most recent five surveys.

The poll graphic below incorporates these changes. The ConHome Poll of Polls now includes BPIX's survey for the Mail on Sunday and Angus Reid's survey for PoliticalBetting.com. Under this refined system we are now projecting a Tory majority of 40+ (ie at least 40).

YouGov

Tim Montgomerie

56 comments for: Two changes to ConHome’s presentation of opinion polls

Leave a Reply

You must be logged in to post a comment.